Example: Binary classification¶

---

This example shows how to use ATOM to solve a binary classification problem. Additonnaly, we'll perform a variety of data cleaning steps to prepare the data for modeling.

The data used is a variation on the Australian weather dataset from Kaggle. You can download it from here. The goal of this dataset is to predict whether or not it will rain tomorrow training a binary classifier on target RainTomorrow.

Load the data¶

In [1]:

# Import packages
import pandas as pd
from atom import ATOMClassifier

# Import packages import pandas as pd from atom import ATOMClassifier

In [2]:

# Load data
X = pd.read_csv("docs_source/examples/datasets/weatherAUS.csv")

# Let's have a look
X.head()

# Load data X = pd.read_csv("docs_source/examples/datasets/weatherAUS.csv") # Let's have a look X.head()

Out[2]:

	Location	MinTemp	MaxTemp	Rainfall	Evaporation	Sunshine	WindGustDir	WindGustSpeed	WindDir9am	WindDir3pm	...	Humidity9am	Humidity3pm	Pressure9am	Pressure3pm	Cloud9am	Cloud3pm	Temp9am	Temp3pm	RainToday	RainTomorrow
0	MelbourneAirport	18.0	26.9	21.4	7.0	8.9	SSE	41.0	W	SSE	...	95.0	54.0	1019.5	1017.0	8.0	5.0	18.5	26.0	Yes	0
1	Adelaide	17.2	23.4	0.0	NaN	NaN	S	41.0	S	WSW	...	59.0	36.0	1015.7	1015.7	NaN	NaN	17.7	21.9	No	0
2	Cairns	18.6	24.6	7.4	3.0	6.1	SSE	54.0	SSE	SE	...	78.0	57.0	1018.7	1016.6	3.0	3.0	20.8	24.1	Yes	0
3	Portland	13.6	16.8	4.2	1.2	0.0	ESE	39.0	ESE	ESE	...	76.0	74.0	1021.4	1020.5	7.0	8.0	15.6	16.0	Yes	1
4	Walpole	16.4	19.9	0.0	NaN	NaN	SE	44.0	SE	SE	...	78.0	70.0	1019.4	1018.9	NaN	NaN	17.4	18.1	No	0

5 rows × 22 columns

Run the pipeline¶

In [3]:

# Call atom using only 5% of the complete dataset (for explanatory purposes)
atom = ATOMClassifier(X, y="RainTomorrow", n_rows=0.05, n_jobs=8, verbose=2)

# Call atom using only 5% of the complete dataset (for explanatory purposes) atom = ATOMClassifier(X, y="RainTomorrow", n_rows=0.05, n_jobs=8, verbose=2)

<< ================== ATOM ================== >>

Configuration ==================== >>
Algorithm task: Binary classification.
Parallel processing with 8 cores.
Parallelization backend: loky

Dataset stats ==================== >>
Shape: (7109, 22)
Train set size: 5688
Test set size: 1421
-------------------------------------
Memory: 1.25 MB
Scaled: False
Missing values: 15868 (10.1%)
Categorical features: 5 (23.8%)
Duplicates: 1 (0.0%)

In [4]:

# Impute missing values
atom.impute(strat_num="median", strat_cat="drop", max_nan_rows=0.8)

# Impute missing values atom.impute(strat_num="median", strat_cat="drop", max_nan_rows=0.8)

Fitting Imputer...
Imputing missing values...
 --> Dropping 7 samples for containing more than 16 missing values.
 --> Imputing 23 missing values with median (11.9) in column MinTemp.
 --> Imputing 10 missing values with median (22.6) in column MaxTemp.
 --> Imputing 72 missing values with median (0.0) in column Rainfall.
 --> Imputing 3059 missing values with median (4.6) in column Evaporation.
 --> Imputing 3382 missing values with median (8.5) in column Sunshine.
 --> Dropping 467 samples due to missing values in column WindGustDir.
 --> Imputing 466 missing values with median (39.0) in column WindGustSpeed.
 --> Dropping 479 samples due to missing values in column WindDir9am.
 --> Dropping 165 samples due to missing values in column WindDir3pm.
 --> Imputing 53 missing values with median (13.0) in column WindSpeed9am.
 --> Imputing 115 missing values with median (17.0) in column WindSpeed3pm.
 --> Imputing 72 missing values with median (70.0) in column Humidity9am.
 --> Imputing 164 missing values with median (52.0) in column Humidity3pm.
 --> Imputing 699 missing values with median (1017.7) in column Pressure9am.
 --> Imputing 699 missing values with median (1015.4) in column Pressure3pm.
 --> Imputing 2698 missing values with median (5.0) in column Cloud9am.
 --> Imputing 2903 missing values with median (5.0) in column Cloud3pm.
 --> Imputing 32 missing values with median (16.7) in column Temp9am.
 --> Imputing 116 missing values with median (21.1) in column Temp3pm.
 --> Dropping 72 samples due to missing values in column RainToday.

In [5]:

# Encode the categorical features
atom.encode(strategy="Target", max_onehot=10, infrequent_to_value=0.04)

# Encode the categorical features atom.encode(strategy="Target", max_onehot=10, infrequent_to_value=0.04)

Fitting Encoder...
Encoding categorical columns...
 --> Target-encoding feature Location. Contains 47 classes.
 --> Target-encoding feature WindGustDir. Contains 16 classes.
 --> Target-encoding feature WindDir9am. Contains 16 classes.
 --> Target-encoding feature WindDir3pm. Contains 16 classes.
 --> Ordinal-encoding feature RainToday. Contains 2 classes.

In [6]:

# Train an Extra-Trees and a Random Forest model
atom.run(models=["ET", "RF"], metric="f1", n_bootstrap=5)

# Train an Extra-Trees and a Random Forest model atom.run(models=["ET", "RF"], metric="f1", n_bootstrap=5)

Training ========================= >>
Models: ET, RF
Metric: f1


Results for ExtraTrees:
Fit ---------------------------------------------
Train evaluation --> f1: 1.0
Test evaluation --> f1: 0.5956
Time elapsed: 1.414s
Bootstrap ---------------------------------------
Evaluation --> f1: 0.5709 ± 0.0198
Time elapsed: 1.020s
-------------------------------------------------
Time: 2.434s


Results for RandomForest:
Fit ---------------------------------------------
Train evaluation --> f1: 1.0
Test evaluation --> f1: 0.6124
Time elapsed: 0.337s
Bootstrap ---------------------------------------
Evaluation --> f1: 0.5802 ± 0.0111
Time elapsed: 1.281s
-------------------------------------------------
Time: 1.618s


Final results ==================== >>
Total time: 4.225s
-------------------------------------
ExtraTrees   --> f1: 0.5709 ± 0.0198 ~
RandomForest --> f1: 0.5802 ± 0.0111 ~ !

Analyze the results¶

In [7]:

# Let's have a look at the final results
atom.results

# Let's have a look at the final results atom.results

Out[7]:

	f1_train	f1_test	time_fit	f1_bootstrap	time_bootstrap	time
ET	0.8503	0.5688	1.414043	0.570892	1.019728	2.433771
RF	0.8552	0.5612	0.336765	0.580178	1.281000	1.617765

In [8]:

# Visualize the bootstrap results
atom.plot_results(title="RF vs ET performance")

# Visualize the bootstrap results atom.plot_results(title="RF vs ET performance")

In [9]:

# Print the results of some common metrics
atom.evaluate()

# Print the results of some common metrics atom.evaluate()

Out[9]:

	accuracy	ap	ba	f1	jaccard	mcc	precision	recall	auc
ET	0.8478	0.6904	0.7059	0.5688	0.3974	0.5108	0.7750	0.4493	0.8561
RF	0.8405	0.6775	0.7038	0.5612	0.3901	0.4891	0.7283	0.4565	0.8502

In [10]:

# The winner attribute calls the best model (atom.winner == atom.rf)
print(f"The winner is the {atom.winner.name} model!!")

# The winner attribute calls the best model (atom.winner == atom.rf) print(f"The winner is the {atom.winner.name} model!!")

The winner is the RF model!!

In [11]:

# Visualize the distribution of predicted probabilities
atom.winner.plot_probabilities()

# Visualize the distribution of predicted probabilities atom.winner.plot_probabilities()

In [12]:

# Compare how different metrics perform for different thresholds
atom.winner.plot_threshold(metric=["f1", "accuracy", "ap"], steps=50)

# Compare how different metrics perform for different thresholds atom.winner.plot_threshold(metric=["f1", "accuracy", "ap"], steps=50)